All-In PodcastTrump's Cabinet, Google's Quantum Chip, Apple's Flop, TikTok, State of VC
CHAPTERS
- 0:00 – 4:00
Cold Open, Banter, and Introducing Guest Host Keith Rabois
The episode opens with lighthearted banter among Jason, Chamath, and Friedberg about skiing, viral clips, and in-jokes before formally introducing investor Keith Rabois as a guest co-host filling in for David Sacks. They briefly reference Rabois’ anti-dictator stance and “founder mode” philosophy as a setup for the deeper discussion.
- •Hosts joke about skiing together, physical quirks, and viral internet clips.
- •Jason introduces Keith Rabois as the stylish, fit stand-in for David Sacks.
- •Early hint that the conversation will cover dictators in politics vs. strong CEOs in business.
- •Mention of Rabois’ conference on ‘Founder Mode’ and his emphasis on strong founder-led companies.
- 4:00 – 13:00
Keith Rabois’ Career Arc and Why He Returned to Khosla Ventures
Rabois recounts his path from PayPal and LinkedIn to Square, Khosla Ventures, and Founders Fund, and why he ultimately moved back to Khosla. He contrasts Khosla’s early-stage, input-driven approach with Founders Fund’s momentum-focused, later-stage strategy and explains why he prefers investing at the “keynote deck” stage.
- •Career highlights: PayPal, LinkedIn, Slide, Square (early employee), then moving into VC in 2013.
- •Explains common pattern of executives becoming VCs via relationships with board members.
- •Left Khosla originally partly due to brutal Sand Hill Road commute; Founders Fund was in San Francisco.
- •Sam Altman’s advice that commute time is inversely correlated with happiness influenced his move.
- •Khosla is early-stage and input-driven; Founders Fund is momentum, later-stage, output-driven.
- •At Khosla he led an unusually late-stage Stripe investment; Anduril and Ramp were similar later exceptions at Founders Fund.
- 13:00 – 23:00
Founder-Mode Investing: Assessing People, Not Metrics
The panel digs into what it takes to write a check off a slide deck, with Rabois emphasizing founder assessment over early metrics. They contrast Khosla’s hands-on, company-building style with Founders Fund’s deliberately ‘hands-off’ posture and discuss how domain expertise can blind incumbents to disruptive ideas.
- •Rabois says his ‘sweet spot’ is pre-product, pre-metrics, based purely on founder quality.
- •Believes metrics at early stage can confuse investors and reduce focus on founder potential.
- •Khosla sees itself as actively increasing a company’s probability of success; Founders Fund prides itself on ‘getting out of the way.’
- •Friedberg shares experiences having both Khosla and Founders Fund as lead investors at different startups.
- •Rabois’ due diligence approach with “experts”: only cares if something is physically impossible; otherwise he often ignores expert skepticism.
- •Domain experts often explain why something can’t work in the old paradigm, missing zero-to-one opportunities like Airbnb, Uber, or PayPal.
- 23:00 – 25:59
Why Keith Isn’t Joining the Administration and Jacob Helberg’s New Role
Rabois explains why he personally declined entering the Trump administration despite strong political interests, while his husband Jacob Helberg has accepted a senior economic role. They frame Helberg’s job as exporting an American economic-strength-first philosophy through foreign policy.
- •Rabois loves politics and worked in it pre-tech, but believes if he leaves VC for 2–5 years he could never effectively return given tech’s pace.
- •Sees himself in the prime of his venture career and wants to see his portfolio companies become public.
- •Plans to consider politics in a decade, not now.
- •Jacob Helberg will serve as Undersecretary of State for Economic Affairs (with other add-ons to the official title).
- •His mandate: build foreign policy grounded in economic strength, viewing economic power as the basis of geopolitical power.
- •They characterize Trump’s first three years as a strong economy and argue for rebuilding American economic strength globally.
- 25:59 – 29:40
Google’s Willow Quantum Chip: Architecture, Error Correction, and Crypto Threats
Friedberg gives a detailed but accessible primer on quantum computing and explains why Google’s Willow chip—demonstrating decreasing error rates as more qubits are combined—is a breakthrough. The group then explores the implications for encryption standards, Bitcoin’s security, and how far we might be from real-world quantum attacks.
- •Quantum bits (qubits) exist as quantum states with superposition and entanglement, unlike classical binary bits.
- •The central challenge is maintaining qubit coherence and suppressing errors long enough to perform meaningful computations.
- •Willow uses arrays of physical qubits to construct logical qubits and shows error rates *declining* as array size grows (3x3 → 5x5 → 7x7).
- •Google claims a benchmark completed in under 5 minutes vs ~10²⁵ years for top classical supercomputers, though verifying correctness is non-trivial.
- •Shor’s algorithm could exploit quantum machines to factor large integers rapidly, threatening RSA and SHA-256 (Bitcoin’s primitive).
- •Chamath estimates ~4,000 logical qubits might break RSA-2048; ~8,000 could threaten Bitcoin’s SHA-256, suggesting a 2–5 year ‘shot clock’ if progress compounds.
- •Rabois cautions that this is likely a decade away for practical use, and many systems are already migrating to post-quantum encryption.
- •Even if quantum arrives, software, languages, and compilers must be reinvented; nothing in today’s stack maps directly.
- 29:40 – 38:20
Trump’s Decision-Making Style and Billionaire Cabinet: Operators vs. Theoreticians
The group analyzes Trump’s emerging cabinet, notable for its high net-worth business figures, and debates whether bringing wealthy operators into government is beneficial or risky. They argue this resembles the founding fathers’ model of temporary civic duty and contrast it with a professional political and bureaucratic class.
- •Rabois describes Trump’s decision style: he canvasses many opinions around a room before deciding, making him hard to predict to the media.
- •Chamath notes this may be the first modern U.S. administration with so many highly successful businesspeople joining government at once.
- •They argue Democrats rarely recruit similarly talented operators because it’s culturally unfashionable on the left to valorize business success.
- •Chamath’s view: the modern economy is too complex to be run by abstract theorists and lifetime bureaucrats.
- •They acknowledge concerns about regulators coming from industry (pharma, ag, energy) but argue transparency and clear rules beat hidden mid-level capture.
- •Rabois suggests an ideal model is successful operators regulating sectors they won’t later work in, mitigating self-dealing while importing competence.
- 38:20 – 41:20
Quantum Weirdness, Many Worlds, and Pop-Science Sidebars
The conversation briefly detours into the philosophical and pop-science aspects of quantum mechanics, including wavefunction collapse, Schrodinger’s cat, and multiverse interpretations mentioned in Google’s materials. While humorous, the discussion reinforces how counterintuitive quantum behavior is compared to classical physics.
- •Friedberg recounts the non-intuitive nature of quantum mechanics courses and how observation collapses quantum states.
- •They touch on the double-slit experiment and how measurement changes whether light behaves as a particle or wave.
- •Discuss the idea that a qubit is in superposition until observed, echoing Schrodinger’s cat thought experiment.
- •Jason cites Google’s comment that Willow lends credence to multiverse interpretations; Friedberg clarifies this is consistent with long-standing quantum theory, not a new discovery.
- •The science segment stays light with jokes but underlines the profound conceptual shift quantum computing demands.
- 41:20 – 43:50
Trump’s ‘Why?’ Superpower and Parallels to Founders
Rabois reflects on Trump’s repeated political resilience despite intense establishment opposition, attributing it to his habit of questioning entrenched processes. They draw a parallel between Trump’s constant ‘why’ and how successful founders challenge industry orthodoxy.
- •Rabois notes the improbability of someone under intense media and legal assault being elected president twice, suggesting Trump has political superpowers.
- •Interviews with ex-Cabinet critics led him to conclude Trump’s main strength is incessantly asking ‘why’ existing policies and practices exist.
- •Many D.C. processes haven’t been reconsidered for decades, making them vulnerable to a ‘why’-driven outsider.
- •This mirrors successful startup founders who enter industries they don’t know, relentlessly question assumptions, and rebuild from first principles.
- •They argue this ‘why’ habit is more predictive of success than domain expertise in many entrepreneurial contexts.
- 43:50 – 54:30
Apple’s AI Server Chips and the Breakdown of Product Quality and ‘Taste’
The hosts react to reports that Apple is building its own AI inference server chips while Jason and Chamath rail against the poor quality of iOS 18 and recent iPhones. Rabois frames Apple as a vertically integrated juggernaut whose loss of singular design leadership, without a compensating data culture, is now showing in flawed user experiences.
- •Apple is reportedly working with Broadcom and TSMC on internal AI inference chips (code-named Baltra), targeting mass production in 2026.
- •These chips are for Apple’s own infrastructure rather than external cloud customers, reinforcing their device-centric, privacy-branded AI strategy.
- •Chamath and Jason describe major iOS 18 regressions: dropped calls, bricked phone functions, and an unusable Photos app and Control Center.
- •Rabois praises vertical integration—down to custom silicon—as an enduring moat (like Steve Jobs envisioned) but notes cultural complacency can follow.
- •He argues Apple’s historical edge came from Jobs/Ive’s high ‘taste’ bar; with them gone and no strong data-driven UX backstop, quality drifts.
- •Unlike Meta/Google, Apple has ‘antibodies’ against data-driven UX, so when taste decays even 10–20%, there’s no scaffolding to catch obvious mistakes.
- •They compare Apple’s situation to organizations losing a visionary coach: decay appears gradually, then suddenly.
- 54:30 – 1:03:55
TikTok Under Fire: Data Access, CCP Law, and Reciprocity
With a January 19 divest-or-ban deadline upheld by an appeals court, the panel dissects why TikTok is viewed as a national security threat. Rabois lays out multiple layers of concern—data access, Chinese law, potential algorithmic manipulation, and asymmetric market access—and suggests classified briefings likely revealed even more alarming details to Congress.
- •TikTok’s CEO previously told Congress U.S. data is stored in Texas and inaccessible to China; subsequent evidence shows ByteDance staff in China *did* access U.S. data.
- •2022 revelations: ByteDance used internal tools to track journalists’ locations to identify leakers.
- •China’s National Intelligence Law obligates companies to share data with the CCP on request, without court oversight—creating an unavoidable structural risk.
- •Rabois argues there’s strong evidence (some classified) that spooked lawmakers, reflected in overwhelming bipartisan votes: 352–65 in the House, 79–18 in the Senate.
- •They also raise reciprocity: U.S. apps like Meta, Google, X, Reddit are blocked in China while TikTok dominates here.
- •Chamath suggests TikTok could host Pegasus-like backdoors: only 1,000–15,000 ‘interesting’ users (officials, elites) matter for surveillance, not 200M Americans.
- •They note CCP’s refusal to allow any divestiture at any price strongly implies TikTok has strategic, not just financial, value for China.
- 1:03:55 – 1:16:40
State of Venture Capital: AI Heat, Crypto Revival, and IPO vs. M&A
The discussion turns to venture markets, where AI and crypto deals are hot while traditional SaaS still struggles with inflated private valuations. Rabois reiterates his pro-IPO stance and critiques Stripe’s prolonged private status, while Friedberg emphasizes that exit bottlenecks stem more from investors’ refusal to accept markdowns than from market conditions or antitrust policy.
- •Deal volumes are returning to pre-COVID levels, but Rabois stresses the market is bifurcated: AI and crypto hot; non-AI enterprise cool.
- •Crypto entrepreneurship picked back up with expectations of a friendlier administration and SEC, after a regulatory chill.
- •Rabois believes meaningful VC fund returns come from IPOs, not small M&A, especially for large $500M–$2B funds.
- •He advocates going public as early as ~$50M revenue with clear line-of-sight to $100–200M, touting accountability and strategic optionality (e.g., equity M&A).
- •Stripe and SpaceX are cited as outliers that have stayed private for decades; he questions Stripe’s reluctance to IPO, warning of disruption risk from stablecoins.
- •Friedberg argues Lina Khan’s antitrust activism is mostly about mega-deals and hasn’t meaningfully blocked typical tech M&A; the bigger issue is investor anchoring on 2021 valuations.
- •Many late-stage investors prefer to ‘earn back’ their marks privately rather than accept 60–70% haircuts in IPOs or acquisitions.
- 1:16:40 – 1:29:10
Tariffs, Immigration, and Industrial Strategy in a New Republican Era
The hosts explore how a Trump administration might use tariffs and immigration enforcement without stoking inflation, contrary to conventional macro fears. They argue that targeted tariffs can work as reverse subsidies that level the playing field against heavily subsidized and environmentally lax Chinese industries, citing examples from EV supply chains, magnets, and lithium.
- •Rabois relays that Wall Street veterans widely respect Trump’s Treasury pick (Scott), seeing him as a sophisticated trader who understands substitution effects and inflation dynamics.
- •He argues Trump’s first-term tariffs did not spike inflation, and internal research (cited by J.D. Vance) blames housing inflation partly on illegal immigration.
- •Chamath sees tariffs as tools to counteract foreign subsidies and environmental arbitrage—e.g., Chinese dominance in permanent magnets and rare earth processing essential for electrification.
- •Tariffs can make domestic production viable when combined with DOE subsidies and strategic industrial policy.
- •They highlight U.S. rare earth and lithium deposits blocked by strict environmental rules; suggest revisiting regulations to reduce dependency on China.
- •Chinese AI models benefit from zero copyright or safety guardrails, training faster and more broadly than U.S. models constrained by IP and policy—a different kind of unfair advantage that may require reciprocal policies.
- •They anticipate tariffs will focus heavily on China while aiming to lower core consumer costs like groceries and housing over time.
- 1:29:10
Crypto’s Speculative Phase, Saylor’s Structure, and Potential Real Utility
In closing, the panel touches on the latest crypto rally and Michael Saylor’s high-leverage Bitcoin strategy via MicroStrategy convertibles. Rabois sees speculation as the main use case today but compares Bitcoin’s network to the Facebook social graph—valuable if someone can build real applications atop it.
- •Rabois believes crypto’s dominant current use case is speculation, similar to retail stock trading or sports betting.
- •He thinks the opportunity is to build utilitarian applications that leverage the already-large crypto ‘graph’ of nodes and holders, akin to apps built on Facebook’s social graph.
- •Friedberg explains Saylor’s MicroStrategy structure as a synthetic call option: investors buy convertibles with coupons plus upside if Bitcoin soars above implied levels.
- •Convertible notes give hedge funds structured exposure: income now, equity-like upside later.
- •The panel agrees crypto markets are highly volatile; long-term value depends on whether genuine productivity layers and apps emerge atop base chains like Bitcoin.
